The concept of "lie detection" is nothing new. The Bedouins used to force witnesses to lick hot iron, thinking that "lying causes a dryness of the mouth" and only liars would have burned tongues. Modern detectors (aka polygraph machines) are slightly more sophisticated, but are lie detectors accurate? What is the science of lie detectors, anyway?
Despite their ubiquity in film and television, and the FBI's insistence on using them as a screening tool, the scientific community agrees there are many reasons lie detectors don't work. As one critic noted, it's "19th-century technology... put in a box." While they may have some value as a prop or deterrent, experts think their continued use is "highly problematic." Why? Read on and don't worry about your heart rate - it doesn't tell us anything.
Unlike the way, say, being afraid can cause your body to react in certain ways, the act of deception doesn't cause your body to exhibit unique physiological signs indicating that you are, in fact, a liar. A polygraph measures heart rate, blood pressure, sweating, and breathing, none of which are linked in any way to lying or truth-telling, according to psychologist Leonard Saxe. (Saxe should know: he authored a 1983 report to Congress that led to a nationwide ban on private employers forcing employees to take polygraph tests.)
The basic theory behind modern polygraph testing - lying manifests itself in your body - is similar to an ancient Chinese trick for interrogating suspected criminals. The interrogator would fill the suspect's mouth with uncooked rice, because liars, the logic went, have dry mouths, meaning if rice stuck to your tongue, you were considered a liar. Polygraphs rely on a similar misconception, made only slightly more sophisticated via a gauntlet of pre-test control questions meant to calibrate what your lie looks like (i.e., no, polygraphy advocates "do not claim that the instrument measures deception directly").
Let's not forget: behind every lie detector is a lie detectorist running the show. The person conducting the polygraph test asks "control questions" that establish what your version of "lying" looks like. What if they're wrong?
Mother Jones reports that a man named George Maschke found out the hard way how an operator, or polygrapher, can ruin test results. Maschke applied to be an FBI special agent but failed because of "deception with regard to each and every relevant question." Since some of those "relevant" questions involved selling narcotics and treasonous acts, Maschke was shocked.
He chalks the failure up to a control question about driving under the influence, something Maschke says he has honestly never done. The polygrapher, Maschke says, assumed that everyone who drinks has driven when they shouldn't have, and "marked" his answer as an example of a lie, screwing up the rest of the results. Maschke went on to create AntiPolygraph.org.
If there was ever a need for a machine to measure anxiety, we wouldn't have to invent it: polygraph machines are quite good at it already. The trouble is, anxiety isn't necessarily associated with veracity.
In fact, frequent liars exhibit less anxious behavior simply because they're better at it. Two convicted federal agents, Robert Hanssen and Aldrich Ames, had no problem breezing through polygraph tests during their long careers. It helps, of course, that they were both double agents spying for the Soviet Union and Russia, meaning they both lied for a living.
Another way of looking at lie detectors is as arousal detectors. Anxiety, fear, deception, and the like are all fairly subjective and hard to measure, but sheer arousal is easily quantified via polygraph.
However, being falsely accused can manifest a similar physiological arousal as being caught dead-to-rights can, meaning lie detectors are actually biased toward innocent individuals, according to Dr. William Iacono, writing in the Journal of Forensic Psychology Practice. Furthermore, the pre-test control questions, Iacono argues, simply can't evoke an emotional impact anywhere near as substantial as being hit with a "relevant" question, largely because most subjects assume the control questions are "trivial." This seriously undermines the basic premise of lie detection.